Thank goodness for analogies. How else would we be able to understand the world of microprocessors and minuscule transistors? Telling some people that future gate oxides on transistors will be 0.8nm, or three atomic layers, thick wouldn’t help them much. Tell those same people that 100,000 of them would have to be stacked on top of each other to achieve the same thickness as a sheet of paper and they’d get the idea.
Intel Corp. and IBM Corp. are among the long list of companies pushing the chip design envelope to ensure that Moore’s law stays in force for the foreseeable future. Intel recently announced a breakthrough allowing it to create, within the next five to 10 years, microprocessors containing more than 400 million transistors and capable of speeds of 10GHz. Today processors run at about 1.5GHz and contain 42 million transistors.
IBM, on the other hand, announced it has launched production of CMOS 9S which will allow the company to build chip circuits as small as 130nm. They will be available in early 2001.
These two companies are by no means alone in their developments, but rather part of a larger move to shrink chips and transistors to their physical limits, at which point the structural limitations of silicon will be reached.
Moore’s law lives
“There are sceptics who were saying that you could not make transistors below say 80 nanometres…[but] I would say the prevailing thought today is that you can certainly make them function at 30 nanometres and some people are saying that 20 is going to be the limit,” said Rob Willoner, market analyst in the technology and manufacturing group with Intel in Santa Clara, Calif.
Moore’s law will be attainable at least until the middle of this decade and possibly to the end of the decade, Willoner added. But Intel is not making a judgement call after that.
IBM agrees. “We also continually find new ways to get around them (the physical limitations of silicon) so in the next five years…there are no fundamental physical limits to continuing along the path of Moore’s law,” said Chuck Henry, senior consulting representative with IBM Canada Ltd. in Ottawa.
Analysts agree the road to smaller chips is fairly smooth for the next few years but add the caveat that getting research to market is never easy.
“The good thing about some of these announcements coming out is that they show that there is some life left in Moore’s law and that we are not going to run into a brick wall in the next couple of years,” said Linley Gwennap, principal analyst at the Linley Group in Mountainview, Calif.
“I think that Intel on their research end always paints a very rosy picture but as we have seen in their daily conduct in the last year or so, the engineering issues – for example bringing all of those signals out – are far from trivial and they don’t know yet how all of that is going to work out,” said Roger Kay, research manager with IDC in Framingham, Mass.
fast CPu, slow computer
For most applications in use today, the slowest link in the process is seldom the CPU. Kay said it is more often the computer’s hard drive access speed and the system BUS that cause computing bottlenecks. Linley agrees but doesn’t let Intel (or other chip manufactures) completely off the hook.
“To a certain extent, I think [claiming the CPU is not the slowest link in the process] is just Intel trying to shift the blame on somebody else. But if [it] is really true (that the processor is not the slow link), and I think to a certain degree it is, then what good is it to have these faster processors?” Linley said.
He quickly answered his own question. “The interesting thing will be to see what applications emerge. One of the things we will see (in the future) is more natural methods of interacting with the computer,” he said. “The power will be there in order to help the computer really interact more as you would interact with a person than you would interact with a machine.”
Henry predicts advantages for the increasingly large world of mobile computing.
“The single most evil thing about electronics is heat, so anything that you can do to reduce the power requirements, reduces heat, which increases reliability, which allows you to increase density, which allows you to use less battery power,” he explained.
A future where one laptop battery can make it from Toronto to Tokyo. Nice.
At some point in the not-too-distant future transistors will be at their natural limit, the point where silicon is shrunk as small as it can go. Long before this, new compounds and solutions will likely be found. But silicon will be with us for some time to come.
“Ten years from today there will be lots of silicon in use,” Henry predicts.
Those interested in checking out the ever-shrinking world of chips can visit Intel’s (www.intel.com/research/silicon) or IBM’s (www.chips.ibm.com) Web sites.
Intel in Rexdale, Ont., can be reached at 1-800-628-8686. IBM in Markham, Ont. can be reached at 1-800-426-4968.